• src/sbbs3/sbbsecho.c

    From Rob Swindell (on Debian L@VERT to Git commit to main/sbbs/m on Tue Oct 27 19:41:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/ad0cde5491bfdcf99dc292ab
    Modified Files:
    src/sbbs3/sbbsecho.c
    Log Message:
    sbbsecho: bundle full packets immediately; keep long runs from going stale

    A remote AreaFix RESCAN of every area could fill the disk: outbound
    packets that reached the maximum packet size were held in the temp
    directory until the end of the run, so a large export accumulated every exported message there, uncompressed (32 GB in the reported case). The ever-growing temp directory and outbound packet list also made each new
    packet slower to create, stretching the run to about 20 hours.

    - Bundle each outbound packet as soon as it is full, and drop it from
    the outbound packet list. pktname() no longer reuses a name within a
    run: a packet already moved into a bundle no longer exists on disk,
    and a same-named packet would replace it inside that bundle.

    - During long runs, refresh the time stamps of sbbsecho.bsy and the BSO
    node lock files (at most once a minute), so a run that outlasts
    BsyTimeout isn't considered stale and joined by a concurrent instance.

    - Check MinFreeDiskSpace for the temp and outbound directories before
    creating each new outbound packet, and stop gracefully (as on a
    termination signal) rather than writing until the disk is full.

    - When an echomail export stops early, save the export pointer of the
    last message processed instead of the sub-board's last message, so
    the remaining messages are exported by the next run rather than
    skipped. This also affected an export interrupted with Ctrl-C.

    Fixes #1301

    Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
  • From Rob Swindell (on Windows@VERT to Git commit to main/sbbs/m on Sat Nov 14 13:33:00 2026
    https://gitlab.synchro.net/main/sbbs/-/commit/114a872aee602dbf30a8719c
    Modified Files:
    src/sbbs3/sbbsecho.c
    Log Message:
    sbbsecho: use the sub-board's storage mode when re-tossing bad echomail

    retoss_bad_echomail() passed "smb.status.attr & SMB_HYPERALLOC" as the
    storage mode to smb_addmsg(), so a message moved out of the bad-echo
    area into a sub-board configured for fast allocation was added with self-packing allocation instead, scanning the data allocation file for
    free blocks on every message. It has done so since the re-toss feature
    was added in 9f2c88e060 (2017-10-29).

    Use smb_storage_mode(), as the normal import path in fmsgtosmsg() does.

    Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>